Semantic polyfunctionality and grammaticalization of the Old English subordinator be þæm þe: A corpus-based study

Matti Kilpiö, Research Unit for Variation, Contacts and Change in English (VARIENG), University of Helsinki
Olga Timofeeva, University of Zurich

1. Introduction

1.1 Dictionary evidence for be þæm þe

The earliest dictionary entry listing be þæm þe dates back to 1888, [1] when the New English Dictionary (NED), vol. I: A–B, was first published. In sense 36.b (s.v. by) be þæm þe along with by that, by reason (that) are classified as ‘the conjunctive phrases’ whose meaning is ‘inasmuch as, because, since’, with the first attestation in the Cotton Homilies dated to around 1175. Moreover, earlier in the same entry (sense 23.d), we find by so (that) ‘if only, provided that’ marked as obsolete, with three citations, all from Langland and dated to around 1393. In the NED, by (OE be) is thus the first element of conjunctive phrases with causal or conditional meanings. The same senses are also found in the Oxford English Dictionary (OED), the historical continuation of the NED.

The Middle English Dictionary (MED) s.v. bī conj. gives the following three categories, ultimately based on semantic criteria:

1. In the temporal clause: by the time that, when; bi than, bi that, bi than that (the);
2. bi so (that): (a) on condition that, if, if only; (b) so that.
3. bi that, bi þam þe, for the reason that, since. – The be þam þe instance cited in this section (from BL Cotton Vespasian A.22) is the same as the Cotton Homily example referred to above in connection with the NED, but it is given a slightly later dating here: a1225 (?OE).

The Dictionary of Old English (DOE), s.v. be, big II, offers essentially the same interpretation as the NED and the OED: ‘[c]onjunction (more strictly, a preposition, the first element of a conjunctive phrase): be þæm þe, rarely be þon þe/be þæm þæt’. The DOE entry is naturally much more detailed both in terms of spelling variants and in terms of context and the semantics of be. Although it makes no explicit distinction between the relative and adverbial subordinator uses of be þæm þe, the former are recorded and exemplified under II.A–B and E, and the latter under II.C–D and F. The adverbial subordinator uses again are provided with interpretations rather than grammatical labels:

II.C.1. according to that which, where þe is complement of the following verb;
II.C.2. according to that which, where þe is object;
II.D. according as, as, where þe is obliquely related to the following verb;
II.F. where þe or þæt introduces a clause explaining þæm: ‘to the extent that, in so far as, in that; because’.

These interpretations, however, allow us to conclude that ‘according to/as’ implying a relation of comparison is one of the most prominent senses in the Old English data. We are going to see shortly that our corpus study corroborates this statistically (see Table 1 below).

1.2 Previous research on be þæm þe

To the best of our knowledge, be þæm þe first features in the historical research into conjunctions and connectives in 1966. Kirsti Kivimaa, in her survey of þe and þat as clause connectives, records a few instances of be þæm þe in Old English under ‘conjunctive phrases’ (1966: 152–153), the overall focus of her study being, however, Early Middle English (ibid.: 168–259). Kivimaa observes that the most frequent sense of be þæm þe is ‘according as’, which she interprets as a relation of manner, comparison or modality. She distinguishes this use from causal, temporal (bi that), and instrumental ones. What is also important is that Kivimaa’s study (based on a select corpus of Early Middle English texts) gives us a glimpse of the diatopic and diachronic distribution of be þæm þe up to 1300. Be þæm þe and its variant be/bi þan/þen þe are still attested in Southern and West and East Midland texts between 1100 and 1200, giving way to be þan þat or more commonly bi þat in Kentish, Southern, West and East Midland texts between 1200 and 1300.

Bruce Mitchell discusses be þæm þe/þæt under ‘grouped prepositional formulae’ and ‘prepositional conjunctions’ that introduce clauses of purpose, cause, and comparison (1987: §§2891, 2917, 3090, and 3368). His account, however, does not go much beyond merely listing instances of be þæm þe among other prepositional connective phrases with similar structure (see also §§2420, 2425–2432).

Although Bernd Kortmann’s 1997 Adverbial Subordination claims to provide ‘the complete inventories of adverbial subordinators’ in Old English (OE), Middle English (ME), Early Modern English (EModE), and Present-Day English (PDE), be þæm þe escapes his attention (1997: ch. 10). His main observation concerning what he classifies as ‘complex subordinators incorporating case-marked demonstrative pronouns’ is that in Old English they constituted almost 19 per cent of the subordinator inventory but that during Late Middle English this category was eventually lost (1997: 292–306).

In his numerous studies on adverbial subordinators, Matti Rissanen records only two occurrences of by the (that) in ME1 (1997: 376), but does not trace it either back to the Old English be þæm þe or later to Early Modern English by that. Rissanen, too, points out that complex adverbial subordinators, ‘formed with a preposition governing the oblique form of the demonstrative noun’ are a frequent and productive type in Old English. He remarks, however, that these subordinators ‘represent various stages of grammaticalisation’, with for þon and mid þy, for example, being more advanced in these terms and more frequent than the others. He thus suggests an ultimate relation between the frequency of a particular subordinator and the degree of its grammaticalization (2007: 182).

Ursula Lenker’s recent monograph on adverbial connectors discusses items similar in structure to be þæm þe under ‘pronominal connectors’, mentioning only be þæm þinge ‘for this reason’ and be þy ‘for this; hence’ among causal connectors (2010: 25, 59, 72–75, 144).

As far as the origin, development and functions of be þam þe in Old English are concerned, this brief survey of available literature shows that, to quote Bruce Mitchell, there is clearly ‘room for more work’.

1.3 Data [2]

To obtain data for this study, we conducted a series of proximity searches in the Dictionary of Old English Corpus (DOEC), selecting the combinations of the preposition be plus demonstrative pronoun in the dative (þæm) or instrumental (þon, þy) plus a subordination marker (þe), i.e., literally ‘by that that’. The following spelling variants were considered:

1. be/bi [3] þam/ðam/þæm/ðæm þe/ðe
2. be/bi þam/ðam/þæm/ðæm þæt/ðæt/þat/ðat
3. be/bi þan/ðan/þæn/ðæn þe/ðe
4. be/bi þan/ðan/þæn/ðæn þæt/ðæt/þat/ðat
5. be/bi þy/ðy þe/ðe
6. be/bi þy/ðy þæt/ðæt/þat/ðat
7. be/bi þon/ðon þe/ðe
8. be/bi þon/ðon þæt/ðæt/þat/ðat

The data was then analysed so as to exclude instances of the relative use of be þæm þe from the general count. The remaining tokens (176), from 95 texts (see Appendix), are thus those that can be unequivocally interpreted as the adverbial subordinator, the great majority (close to 90 per cent) representing spelling type (1). The instances of the adverbial subordinator be þæm þe can be classified into seven semantic types (as shown in Table 1).

We are now going to have a close look at these types and their properties.

2. Analysis

According to Kortmann’s classification of adverbial subordinators (1997: 80–81), the interclausal adverbial relations in Table 1 fall into two major domains: the CCCC domain (causal, conditional, concessive, and contrastive relations) [4] includes be þæm þe of degree, condition, cause, purpose, and result, while the modal domain includes be þæm þe of comment, and similarity and comparison. We begin this analysis with the CCCC, proceeding from the most to the least frequent semantic relations, and then go on to the modal.

2.1 The CCCC domain

2.1.1 Clauses of degree

Kortmann uses the following expression to define degree: ‘Degree/Extent (degree): “q, to the extent that p”’ (1997: 87), where ‘p stands for the proposition expressed by the adverbial clause and q for the proposition expressed by the matrix clause’ (ibid.: 84). Thus, ‘p qualifies the extent to which q holds true or is justified’ (ibid.: 87), the ‘ideal’ adverbial subordinator for this relation being insofar as (ibid.: 80–81). degree is illustrated in Present-Day English by the following two examples (ibid.: 87):

(1)

Their father is also guilty, inasmuch as he knew that his sons were planning to rob a bank;

(2)

We agreed with the lecturer (only) insofar as she condemned England’s foreign policy.

It is this interclausal relation that be þæm þe most often marks in Old English (70 per cent of be þæm þe tokens):

(3)

Be þam ðe he soð God is, he wæs symle acenned of ðam ælmihtigan Fæder …; be þam þe he mann wæs, he wæs of Marian acenned; be þam þe he lichama wæs, he læg bebyrged (ÆHom 10.151)

‘Insofar as he is the true God, he was always born of the almighty Father …; insofar as he was a man, he was born of Mary; insofar as he was a body, he lay buried’ [5]

(4)

Gyf he his lif misfadige, wanige his wyrðscipe be ðam þe seo dæd sy (WCan 1.1.2 (Fowler) 68)

‘If he live a wrongful life, let his honour diminish, to the extent that the deed be’

In example (3) we see that the nature of Christ is defined by means of three propositions that hold true to the extent of the three corresponding parts of the divine entity – ‘born of God, to the extent that he himself is God’, etc. Similarly in example (4), the punishment for wrongful actions should be determined in accordance with their gravity.

2.1.2 Condition

One very interesting development, not mentioned in Mitchell 1987, the DOE, or elsewhere in the literature, is the Late Old English rise in the use of be þæm þe in conditional clauses. Select corpus evidence is given here; the path of grammaticalization that we assume to have taken place from ‘degree’ to ‘condition’ will be discussed below in section 2.8.1. There are eleven instances altogether, six typical examples being listed here:

(5)

Þy we beodað Godes bebode & eallra his haligra, þæt nan þara cristenra manna þe þis gehyre him beforan rædan oððe elles hwara hit him gereccan, ne geþristlæce he þis fæsten to abrecenne, be þam þe he wille him for Gode geborgen habban (HomS 34 (ScraggVerc) 19 99)

‘Therefore we command with the command of God and all his saints, that no Christian who hears this read before him or elsewhere hears it being interpreted to him, that he must not presume to break this fast if he wishes to be protected before God.’

(6)

And we lærað þæt ænig gehadod man his scare ne helige, ne hine misefesian læte, ne his beard ænige hwile hæbbe, be þam þe he wille Godes bletsunge habban and sancte Petres and ure (WCan 1.1.1 (Fowler) 47)

‘And we enjoin that no ordained man cover his tonsure nor let his hair be cut amiss, nor wear a beard at any time if he wishes to receive blessings from God, St. Peter and us.’

(7)

Riht is, ðæt preostas and efen wel nunnan regollice libban and clænnysse healdan be þam, [6] þe hi willan on mynstran gewunian oððon for worulde weorðscypes wealdan (WPol 2.1.1 (Jost) 186)

‘It is right that priests and also nuns live according to the rule and observe cleanness if they wish to live in the monastery or wield secular power.’

(8)

And se cyngc beodeð eallum his gerefan on æghwilcere stowe, þæt ge þam abbodan æt eallum worldneodum beorgan, swa ge betst magon, & be þam þe ge willan Godes oððe minne freondscipe habban, filstan heora wicneran æghwar to rihte, þæt heo sylfe magan þe oftor on mynstrum fæste gewunian & regollice libban (Law VIIIAtr 32)

‘And the king commands all his reeves at all places that you protect the abbot in all temporal needs as best as you can, and if you wish to have God’s friendship or mine, you shall support their stewards in every respect in what is right, so that they themselves can the more often live securely in monasteries and live according to the rule.’

(9)

Swa þæt nan þæra cyninga ðe cumað æfter me oððe ealdorman oððe oðer rica mid ænigum riccetere oððe unrihte þiss ne awende odde gewanige be þam þe he nelle habban Godes awyrgednysse & his halgena & minne & minra yldrena þe þas ðing fore synd gefreode on ecum freote on ecnysse (Ch 779 (Rob 48) 56)

‘So that none of the kings who succeed me, or an ealdorman or another rich person may change this with arrogance or injustice or annul it if he does not want to be cursed by God, his saints and me and my ancestors, for whose sake these things have been liberated in everlasting freedom in eternity.’

(10)

And riht is þæt ælc man leornige þæt he cunne pater noster and credan, be ðam þe he wille on gehalgodum licgan oþðon husles wyrðe beon, forðam he ne bið wel cristen þe þæt geleornian nele, ne he nah mid rihte oðres mannes to onfonne æt fulluhte ne æt bisceopes handa, se ðe þæt ne can, ær he hit geleornige (WCan 1.1.2 (Fowler) 22)

‘And it is right that each man study so that he know the Pater noster and the Creed, if he wishes to lie in consecrated ground or to be worthy of the Eucharist, because he is not properly a Christian who does not want to learn that; neither has he the right to receive another person at baptism from the bishop’s hand, who does not know that, before he learns it.’

All the instances of conditional be þæm þe clauses express an open condition (cf. Quirk et al. 1985: §15.35) and have a dynamic verb phrase consisting of willan (nyllan) + infinitive. The subject of the conditional clause is animate and personal. All in all, the eleven instances form a uniform group semantically and syntactically.

2.1.3 Cause

There are two instances of be þæm þe/þæt marking a causal relation, both of them from Ælfric:

(11)

Hit gewearð swa be þam þæt him wann on Penda, Myrcena cyning, þe æt his mæges slege ær Eadwines cyninges <Ceadwallan> fylste; and se Penda ne cuðe be Criste nan þincg, and eall Myrcena folc wæs ungefullod þa git (ÆLS (Oswald) 150)

‘It happened so because Penda, king of the Mercians, fought against him, Penda, who assisted Cadwalla in the murder of his relative king Edwin; and this Penda did not know anything about Christ and the whole Mercian nation still remained unbaptized.’

(12)

& se ræd wæs æfre on his rædfæstum geþance, þæt he wircan wolde þa wundorlican gesceafta, be þan ðe he wolde þurh his micclan wisdom þa gesceafta gescippan & þurh his soðan lufe hig liffæstan on þam life, þe hig habbað (ÆLet 4 (SigeweardZ) 29)

‘and in his wise thought there was always the plan that he would create the wonderful creations, because he wanted through his great wisdom to make the created beings and in his true love to endow them with the life which they have.’

The causal use of be þæm þe/þæt is very rare but is attested (for earlier discussions of the variant with þæt in the literature, see Mitchell 1987: §3090). The causal sense could be seen to arise from the polysemy of the preposition be: the DOE, s.v. be, big prep., conj. and adv., I.F.5, lists examples of the causal use of the preposition. For a discussion of the relationship between the semantics of the preposition as compared to the semantics of the subordinator, see section 2.4 below.

It is worth noting that example (11) contains the only recorded example in Ælfric of the use of þæt as the final element in the subordinator introduced by be þæm, since in all the other Ælfrician instances the last element is þe. The status of þæt in the prepositional phrase be þæm þæt will be discussed below in section 2.6. All three examples express a direct reason relationship (cf. Quirk et al. 1985: §15.45).

A parallel to the causal use of be þæm þe/þæt is provided by the Old High German (OHG) prepositional phrase bī thiu, which is lexically, syntactically and semantically related to OE be þy and be þæm. When bī thiu is used as a subordinator it expresses the semantic relation of cause (Wolfrum 1960: 230ff.), as in the following example from the OHG Tatian:

(13)

murmurotun tho thie Iudei fon imo bithiu her quad: ih bim thaz brot ...murmurabant ergo Iudaei de illo quia dixisset: ego sum panis... (Tatian: John 6: 41-42; [Wolfrum 1960: 229])

‘Then the Jews murmured concerning him because he said: I am the bread...’

2.1.4 Purpose and result

Just as in Present-Day English (cf. Quirk et al. 1972: §11.40), Old English clauses of purpose and result overlap in meaning and are usefully discussed together. The corpus contains three instances of clauses of purpose and one clause of result. The following examples illustrate these two relations:

(14)

Ic þe mæg eac reccan sum bispell be þæm þæt þu hit meaht þe sweotolor ongitan, ðeah <hit> ungesceadwise men ongitan ne mægen (Bo 39.132.4)

‘I can also tell you an example so that you can understand it the more clearly, although irrational people are not able to understand it.’

(15)

Nu cweðað sume men þe ðis gescead ne cunnon þæt se mona hine wende be ðan ðe hit wedrian sceall on ðam monðe, ac <hine> ne went næfre naðor, ne weder ne unweder, of ðam ðe his gecynde is (ÆTemp 8.11)

‘Now, some people who do not know this distinction say that the moon changed itself so that there will be a change of weather during that month, but it changes neither, either good or bad weather, from what is its proper nature.’

Example (14) and another Boethius instance not cited here have þæt as the last element of the subordinator while the rest have þe. The occurrence of þæt as the last element is striking but not without parallels in other prepositional subordinators occurring in clauses of purpose. Mitchell (1987: §2889) points out that in prepositional subordinators occurring in clauses of purpose and result þæt rather than þe occurs as the last element. Rissanen (forthcoming), in discussing the subordinator to þon þæt indicating purpose, says that ‘the subordinate clause is perhaps analysed as a kind of noun clause which demands the linking þæt.’ We shall return to the syntactic interpretation of the element þæt below in section 2.6.

The use of be þæm þe to introduce a clause of result (example 15) has not been recognized in earlier studies. This interpretation does not receive support from any Latin source, being entirely based on the context: the moon changes and this is supposed to produce a change of weather as a result.

2.2 The modal domain

2.2.1 Comment clauses

It has been observed by several scholars that adverbial finite clauses are often used as comments or parentheticals (Quirk et al. 1985: §§15.53–55; Brinton 2008: 4–18). Kortmann’s gloss for this kind of interclausal relation is ‘as p, q’, which is, he admits, somewhat evasive. The main idea behind this gloss is that ‘p expresses the speaker’s comment on the content of the matrix clause, typically with the aim of affirming the truth (and thus reliability) of q’ (Kortmann 1997: 87). The comment can contain ‘the source of the speaker’s information, or express agreement with somebody else’s opinion’ (ibid.: 88). Its verb phrase, therefore, typically includes a verb of cognition or utterance. In our data, comment clauses come second in terms of frequency (16.5 per cent of be þæm þe tokens).

(16)

Ða geseah ðæt wif ðæt ðæt treow wæs god to etenne, be ðam ðe hyre ðuhte, & wlitig on eagum & lustbære on gesyhðe (Gen 3.6)

‘Then the woman saw that the tree was good to eat, as it seemed to her, and beautiful to the eyes and desirable to the sight …’

(17)

Nis nan leodscipe swa grædig goldes and seolfres swa þa Judeiscean and þa Romaniscean be þam þe lareowas on bocum awriton (ÆAdmon 1.9.33)

‘No nation is so greedy for gold and silver as the Jews and the Romans, as the teachers have written in books’

(18)

Ymnes and capitle, readinge and fers & halsunga syn anum gemete gehealdene eallum dægum, be þam þe we ær cweðon (BenRW 18.57.4)

‘Hymns and lessons, readings and verse and exorcisms should be held in the same fashion all days, as we have said earlier’

With a verb of cognition (as in example 16), be þæm þe comment clauses express tentativeness over the truth value of the main clause. The fruits of the tree seemed delicious and beautiful to Eve, which may or may not imply that they were really delicious and beautiful. With verbs of utterance (as in examples (17) and (18)), comment clauses are affirmative, implying the truth of the main clause (cf. Quirk et al. 1985: §15.55; see also section 2.3 below).

2.2.2 Similarity and comparison

The last and least frequent type of be þæm þe clause conveying the general notion of comparison (cf. sections 2.1.1 on clauses of degree and 2.2.1 on comment clauses) consists of clauses of similarity and comparison. There are six instances of these in our corpus, five of them from works by Ælfric.

(19)

& he eac þa gesette mid gehadodum preostum, be þam þe se Halga Gast him simble gewissode, swa swa us segð seo racu (ÆLet 4 (SigeweardZ) 1041)

‘and he also provided these churches with ordained priests, as the Holy Spirit always instructed him, as the narrative tells us.’

(20)

Cunnian þonne bealdlice hu wel him licie þeo byrnende þelu & þa hatan tindas on to ridenne be þam þe him ær licodon oðra manna hengestas to stelenenne (WCan 1.2 (Torkar) 45)

‘Let them then boldly check how pleasing the burning board and the hot prongs are for them to ride on, as compared to how it pleased them formerly to steal other people’s horses.’

(21)

Gif se lareow wel tæce. & yfele bysnige: doð swa swa he tæhð. & na be ðam þe he bysnað (ÆCHom I, 17 315.72)

‘If the teacher teaches well and gives a bad example, do as he teaches and not as he instructs by example.’

(22)

Be ðisum is oft micel embspræc, þonne ða læwedan wyllað habban ðone monan, be ðan þe hi hine geseoð, & ða gelæredan hine healdað be ðisum foresædan gesceade (ÆTemp 8.5)

‘About this there is often a lot of discussion, when the lay people want to consider the moon as they see it and the learned ones regard it according to the aforesaid distinction.’

A manner sense is approximated in some of these instances (see, e.g., example 19); in fact, in Quirk et al. (1972: §11.41) clauses of similarity are called clauses of manner. Discussing Present-Day English clauses of similarity and comparison, Quirk et al. (1985: §15.50) point out that similarity is combined with manner when the verb is dynamic. With the exception of example (20), and possibly also example (22), the verb in the Old English examples here is dynamic.

Be þæm þe is clearly a minority variant as a subordinator of clauses of comparison, since swa used singly or in combination reigns supreme in this function (see Mitchell 1987: §§3262–3322). It is interesting to see comparative swa swa and be þæm þe occurring side by side in example (21), suggesting that Ælfric may have used be þæm þe as a stylistic variant here. For the use of the preposition be indicating comparison, see section 2.4 below.

2.3 Collocations

The study of the syntagmatic level, including collocations, [7] is particularly relevant with a subordinator like be þæm þe, which has very little intrinsic meaning in itself. In the final analysis, the co-text provides the key for the semantic classification of the examples. Over 70 per cent of all be þæm þe tokens occur in collocations. In Table 2 we list clauses that are found at least twice in our data.

The collocations in our data fall into three groups. In the first group, the collocates of be þæm þe belong to a domain that can be defined very broadly as that of ‘deeds, actions’ whether past, stereotypical or potential. The typical collocates include: dæd ‘deed, action’ (as in example 4), don ‘to do, make’, earnian ‘to labour for; deserve’, gewurðan ‘to happen to somebody, to be somebody’s due’, and wyrcan ‘to work, make, perform’. The noun dæd is particularly frequent in legal texts, and normally implies ‘according as the deed may be’ in collocations with be þæm þe, referring to a punishment, remuneration or reward given in accordance with a crime, offence or service. Similarly, collocations with the four verbs in this group imply ‘according to what somebody did/deserved’, thus also establishing a degree relation between a deed, misdeed or sin and its remuneration on earth or in heaven. These collocations occur mostly in laws and sections of homilies or sermons dealing with divine punishment and are thus clearly text-type specific.

The second group can again be described very broadly as ‘modal’. Here we find verbs and nouns that establish a relation of degree or extent between the main proposition and somebody’s strength, power, will or need: mæð ‘measure, degree’, miht ‘might, power’, magan ‘to be strong, to be able, may’, neod ‘need, necessity’, onhagian ‘to be within a person’s power or means’, gerædan ‘to determine, decide’ and willan ‘to will, wish, want’. The collocations occur mostly in the didactic rather than narrative sections of homilies and sermons.

While the first two groups of collocations generally express interclausal relations of degree, the third group occurs only in comment clauses. The most typical collocates of be þæm þe here are awritan ‘to write, describe’, cweðan/secgan ‘to say, speak’, and læran/tæcan ‘to teach, instruct, show’. [8] Although semantically this group looks quite uniform, consisting of verbs of utterance (whether oral or written), functionally there are two distinct sub-groups. The first sub-group of collocations works more in terms of pragmatics, invoking the authority of a saint, church father, or the Scripture: ‘as the Book teaches us’ or ‘as St. Augustine says’. These collocations are thus meant to establish the validity of a new text in relation to its transmission of the orthodox truths. [9] The second sub-group works more at the level of discourse, ensuring cohesion between paragraphs: ‘as I/we have said before’. These clauses ensure a smooth flow of utterance, particularly if introducing a recapitulation of exactly what has been said before, making it easy to follow. Both sub-groups occur in homilies/sermons, where recourse to authority and repetition could be used advantageously for the benefit of the audience.

2.4 The polysemy of the preposition be and its possible connection with the polysemy of be þæm þe

One of the crucial questions to be taken into account when addressing the semantic polyfunctionality of the subordinator be þæm þe is the extent to which this polyfunctionality is the result of grammaticalization processes taking place within the subordinator or how far it is a reflection of the attested polyfunctionality of the preposition be.

In the preceding discussion we have established seven types of adverbial relation for the subordinator be þæm þe: degree, comment, similarity and comparison, cause, purpose, result, and condition. Of these relations, the following three are attested for the preposition be in the DOE s.v. be, big prep., conj. and adv.:

• degree (I.D.2.d), as in

  Solil 3 66.10: ac ælc hefð be hys gearnunga swa wite swa wuldor

  ‘But each will receive in proportion to his deserts either punishment or glory’ [translations added by us here and in the following two citations from the DOE]

• similarity and comparison (I.D.2.e), as in

  Or 2 6.50.25 … hwelc gewinn þa wæron be ðæm þe nu sindon …

  ‘what kind of fights those were in comparison with those which now take place…’

• cause (I.F.5), as in

  Bo 38.119.28 ða cwæð he be hwæm cwest þu þæt?

  ‘Then he said, “For what reason do you say that?”’

Thus nearly half of the seven types of meaning of the subordinator are attested for the preposition. Conversely, some common senses of the preposition, such as locative (position or motion) and temporal, are not reflected in the senses of the subordinator. By way of contrast, in Gothic the temporal sense of the preposition bi, cognate with OE be, is reflected in the temporal subordinator biþê ‘while, when’, which has the neuter instrumental þê as the second element (Braune & Ebbinghaus 1973: §§153, 218).

Of the shared meanings, those of similarity and comparison and cause attest to a direct semantic link between preposition and subordinator, but the paucity of instances of causal and comparative be þæm þe makes it difficult to substantiate this claim. With the instances expressing the adverbial relation of degree, the situation is different in that the sense ‘according as’ of be and set phrases branching from it form a natural link between the preposition and the predominant ‘according as’ sense of the subordinator.

2.5 The syntactic polyfunctionality of be þæm þe/þæt

Although this study is primarily concerned with the semantic polyfunctionality of be þæm þe/þæt, a brief discussion of the syntactic polyfunctionality of this sequence is in order. In addition to its use as a pronominal subordinator in adverbial clauses, it is also used (a) in relative constructions and (b) in constructions in which the final element, þe or þæt, introduces a nominal clause. Here are two examples of this use in relative constructions:

(23)

Đære sawle wlyte is þæt heo wisdom lufie; na ðone eorðlican wisdom be þam þe þus awriten is, sapientia huius mundi stultitia est apud deum: þysses middaneardes wisdom is stuntnis ætforan gode (ÆLS (Christmas) 227)

‘It is an adornment to the soul that it loves wisdom; not the worldly wisdom about which it is thus written: sapientia huius mundi stultitia est apud deum: the wisdom of this world is folly in the eyes of God.’

(24)

Hig cwædon eft to þam blindan, hwæt segst þu be þam þe þine eagan untynde; He cwæð, he is witega (Jn (WSCp) 9.17)

‘They said again to the blind man: “What do you say about him who opened your eyes?” He said: “He is a prophet.”’

In (23) the relative particle þe is optional and the demonstrative pronoun þam is anaphoric, referring back to the masculine noun wisdom. In (24), the reference of þam is cataphoric and þe is an obligatory subject introducing a relative clause.

Examples (25) and (26) illustrate the construction in which the final element introduces a nominal clause.

(25)

BE ĐAM ĐÆT ÆLC OĐRUM HYRSUMIGE (BenR 71.130.10)

‘Concerning [the injunction] that each one should be obedient to the other.’

(26)

Be ðam þe se dædbeta ne healt þæt his scrift him tæcð (Conf 3.1.1 (Raith Y) 1.8)

‘Concerning [the possibility] that the penitent does not do what his confessor prescribes for him.’

Both the instances cited, (25) and (26), occur in headings. The demonstrative ðam, which can only be translated rather freely, is cataphoric in both examples. There is no doubt about the nominal clause subordinator status of ðæt in (25), but Mitchell is sceptical about the possibility of þe being a subordinator introducing a nominal clause (1987: §1957). Our interpretation of þe in (25) as introducing a nominal clause is, however, supported by five similar instances, all of them having be ðon þe in headings of the Alfred-Ine law code (LawAfRB 7; LawIneRb 29, 30, 31 and 47).

The syntactic polyfunctionality discussed here is not entirely trivial when it comes to disambiguating the adverbial subordinator from the types of be þæm þe/þæt sequence exemplified by examples (23), (24), (25) and (26).

Consider examples (27) and (28):

(27)

ac manna gehwylc oðrum beode þæt riht, þæt he wille, þæt man him beode, be þam þe hit mæð sy (HomU 59 (Nap 37) 21)

‘but let each man offer to the other one the same equity that he wishes to be offered to him, according as his measure be.’

(28)

& his hiredcnihton eallon V pund to gedale ælcon be þam þe his mæð wære (Ch 1492 (Nap-Steven 10) 25)

‘and to all his domestics five pounds to be shared by each one according as his measure would be/according to that which would be his measure.’

In example (27), there is no ambiguity concerning the adverbial subordinator status of be þam þe as it is followed by hit functioning as a formal subject. The situation is different with example (28): we have opted here for a reading in which be þam þe is a unit representing the subordinator but it would also be possible to interpret be þam as a prepositional phrase containing the cataphoric demonstrative pronoun þam, followed by a relative clause introduced by þe.

2.6 The last element of the subordinator: þe or þæt

There are only three instances of þæt as the subordination marker in our material, the remaining 173 instances having þe. This disproportional breakdown is in harmony with what Rissanen has found about the relative frequency of these two elements in other prepositional subordinators. According to Rissanen (2007: 190 and Rissanen forthcoming), þæt is uncommon as a subordination marker in a number of adverbial subordinators he has studied. Here are the proportions of þe and þæt in five adverbial subordinators in his Helsinki Corpus material:

This suggests that of the two main variants of the subordinator formed with be + dem.pron. + þe/þæt only the predominant variant with þe achieved the status of an established subordinator in Old English.

The origin of the subordinating particles þe and þæt has attracted a lot of attention in the literature (see, e.g., Mitchell 1987: §2427). Kivimaa (1966: 162) suggests that þe is relatival in be þæm þe he wille; thus the sequence be þæm þe could be translated not only by ‘according as he wishes’ but also ‘according to what he wishes’. Compare this to the ambiguity of example (28). As to the origin of the particle þæt, it has been suggested that it originally introduced a nominal clause, at least for purpose clauses (see the citation from Rissanen forthcoming in section 2.1.4).

2.7 Reasons for the preponderance of be þæm þe/þæt in clauses of degree

Why is the semantic relation of degree by far the most common one among the seven attested semantic relations? Any answer to this question must remain tentative, since fully conclusive answers would require systematic study of the inventories of subordinators expressing all seven relations. Still, it is possible to offer hypotheses for the attested state of things.

Other subordinators which offer overpowering competition are available for most of the semantic relations expressed by be þæm þe/þæt. These include for þæm þe (causal clauses), swa (swa) (comment clauses; clauses of similarity and comparison), þæt(te), swa þæt(te) (clauses of purpose and result) and gif (conditional clauses), all of them extensively researched high-frequency items.

For the degree relation there are, in addition to be þæm þe/þæt, a number of alternative subordinators. Mitchell’s discussion of clauses of comparison (1987: §§3202-3385) proved useful in the process of recognizing some of them. The analysis of these ‘competing’ degree subordinators took place in two phases:

• Three subordinators, þæs þe, be þæm dæle þe/þæt and æfter þæm þe were studied within the framework of the DOEC by analysing the entire data and retrieving the degree instances: þæs þe yielded 34, be þæm dæle þe/þæt 13, and æfter þæm þe 6 such instances. All these figures are considerably below the total of 124, of be þæm þe/þæt expressing the degree relation in the DOEC.
• The remaining analyses were carried out in a pilot study by using the syntactically tagged YCOE. The CorpusSearch program was used to retrieve subordinate clauses introduced by swa, combinations involving swa, the univerbated ealswa and þæs without the appended þe. The YCOE was chosen because swa and þæs are high-frequency items. Swa and the combinations involving swa yielded eight instances altogether expressing the degree relation, ealswa and þæs none. Considering the fact that the YCOE yields 24 degree instances of be þæm þe/þæt, it is clear that even swa (+ possible combinations) does not offer real competition for be þæm þe/þæt in expressing the degree relation (see section 2.1.1).

All in all, this tentative analysis of the competition be þæm þe/þæt had to face from alternative subordinators in the various semantic relations discussed in this essay points to the conclusion that be þæm þe/þæt obviously had an opportunity to assert itself in expressing the semantic relation of degree rather than in any other semantic relation. This is directly reflected in the preponderance of this subordinator in the degree relation.

2.8 The rise and grammaticalization of be þæm þe/þæt

In this section, it is our task to try to account first of all for the rise in the subordinator be þæm þe/þæt and for the semantic polyfunctionality of be þæm þe/þæt in terms of the grammaticalization processes.

We understand grammaticalization as being ‘concerned with such questions as how lexical items and constructions come in certain linguistic contexts to serve grammatical functions or how grammatical items develop new grammatical functions’ (Hopper & Traugott 2003: 1). Grammaticalization is linked up with reanalysis and often with an analogical extension of existing patterns (see, e.g., Salminen 2002: 303–305 and Itkonen 2002: 416–418). In the case of be þæm þe/þæt, reanalysis consists in the reinterpretation of the presumed original clause-introducing functions of þe and þæt as semantically vacuous subordinating particles which, when appended to the prepositional phrase be þæm, form a unitary sequence with the phrase and enter the category of pronominal subordinators. Analogical extension of the pattern provided by existing subordinators with pronominal deixis, such as for þæm þe or ær þæm þe, can plausibly be seen as the chief driving force behind the rise of the subordinator be þæm þe/þæt.

The texts that represent Early Old English in our corpus are few: three law codes (LawIIAs, LawVIAs and LawIIEw) and one philosophical text (Alfred’s Boethius translation). The evidence they provide is scanty, seven instances in all, but may give some indication of the early stages of the introduction of be þæm þe/þæt. It is noteworthy that all the five instances of the subordinator in the early law codes represent the adverbial relation of degree, a relation well represented among the instances found in Late Old English secular and ecclesiastical legislation and monastic rules as well as in other text types. Could the early degree instances of be þæm þe be considered as spearheading the rise of the subordinator in general and as pointers towards the later predominance of the degree relation in our corpus?

It was suggested in section 2.4 that the relations of degree, cause, and similarity and comparison expressed by the subordinator are linked to the corresponding senses of the preposition be. What is particularly interesting in these correspondences is that the degree of grammaticalization seen in them is high since degree and cause represent the CCC domain (causal, conditional, concessive, and related interclausal relations) of Kortmann 1997, only the third relation, similarity and comparison, belonging to Kortmann’s modal category.

In the light of the present evidence, the rise of the remaining relations, those of purpose, result, condition, and comment will be seen as the result of grammaticalization processes having taken place after the rise of the subordinator as a unit. The fact that the purpose relation is found in two Boethius instances in Early Old English suggests that this relation (and, logically, its mirror image, the result relation, albeit with no Early Old English attestation in the data) might have been the first to appear. As opposed to this, the relation of condition appears to be the temporally last and final grammaticalization process undergone by be þæm þe, as we shall argue below.

Clauses of purpose, result, and condition all belong to Kortmann’s CCC domain, clauses of comment to Kortmann’s modal domain.

To sum up, five of the seven functions of be þæm þe/þæt already represent the CCCC relations, the network towards which locative, temporal and modal relations move. Only two functions, those represented by comment clauses and clauses of similarity and comparison, represent the modal relations. This suggests that the degree of grammaticalization seen in be þæm þe/þæt is high.

2.8.1 The grammaticalization process of be þæm þe from a subordinator of degree clauses into a subordinator of conditional clauses

According to Kortmann (1997: 196), CCCC relations constitute ‘the prototypical “goal network” within the semantic space of interclausal relations’. Thus, when adverbial subordinators undergo semantic change, the direction is towards the CCCC group, with the conditional domain, in particular, exercising a strong pull, which tends to increase the inventory of conditional subordinators (197–204).

We see the rise of the conditional use of be þæm þe as a result of grammaticalization. The obvious starting-point is the predominant ‘according as, to the extent that’ sense of this connector. The movement is thus from degree to condition. The development seems a natural one. Clauses of degree (a subclass of clauses of proportion) express a proportionality or equivalence of tendency or degree between two situations formalized by Kortmann (1997: 87) as ‘q, to the extent that p’. In conditional clauses, the situation in the matrix clause is contingent on that in the subordinate clause: ‘if p, then q’ (Kortmann 1997: 85). According to Quirk et al. (1985: §15.30), it is a small step from the proportionality seen in clauses of proportion to the contingent relationship seen in conditional clauses. As will be seen in section 2.8.3, there are parallels for the change degree → condition in other languages, such as Finnish and German.

2.8.2 Univerbation?

There are nine examples of beþam (spelt as one word) þe in our data. The majority of these examples (as has often been the case before) – seven – are found in Ælfrician texts. The spelling variant beþam þe introduces adverbial clauses of degree (8 tokens) and comment (1 token), exemplified by examples (29) and (30) respectively.

(29)

Do gehwa georne on godes est, beþam þe hine fyrmest onhagie (HomS 13 (Ass 11) 84)

‘Let everyone act earnestly according to God’s will, insofar as his ability best allows’

(30)

Ic Ælfric abbod on ðisum Engliscum gewrite freondlice grete mid Godes gretinge Wulfget æt Ylmandune. Beþam þe wit nu her spræcon be ðam Engliscum gewritum, ðe ic þe alænde, þæt þe wel licode þæra gewrita andgit, and ic sæde, þæt ic wolde þe sum asendan git (ÆLet 6 (Wulfgeat) 1–3)

‘I, abbot Ælfric, in this English letter, greet Wulfgeat at Ylmandun in a friendly manner, with God’s greeting. As we have both said about these English writings that I lent you, that you were well pleased with their contents, and I said that I wanted to send some [part of them] to you’

On the one hand, the fact that these spellings are only attested within degree and comment examples correlates significantly with the high frequency of degree and comment be þæm þe elsewhere in our data and might be indicative of univerbation of the preposition and demonstrative elements, which is a well-documented phenomenon in the history of the prepositional-pronominal adverbial subordinators in English and other Germanic languages (see Kortmann 1997: 259-264; Lenker 2007: 215; cf. Rissanen 2007: 174). [10] Moreover, our count of one-word spellings relies only on the DOEC, and may exclude a number of instances whose spellings have been consciously changed by the editors. On the other hand, the proportion of one-word beþam spellings is rather low (about 5 per cent of the data) and does not allow us to see any dynamics except for the absence of these spellings in Early Old English. The association between the majority of beþam instances and one particular author makes any conclusion still more tentative. We therefore abstain from final judgement on the univerbation of beþam before data from Early Middle English is investigated more carefully.

2.8.3 German and Finnish parallels of degree → condition cline

Although seemingly unrecorded in the scholarly literature, the development from degree adverbial meaning to conditional is not limited to Late Old English subordinators. A similar change can be observed in modern German. According to the Grimms’ Deutsches Wörterbuch, sofern was both a degree and conditional subordinator around 1854, when the dictionary was first published (Grimm & Grimm [1854] 1905: s.v.sofern):

(a)

sofern, mit deutlicher hervorhebung des begriffs von fern, in der bedeutung so weit als, in dem masse dass

‘sofern with a clear emphasis on the concept of ‘far’ in the sense ‘as far as’, ‘to the extent that’

(b)

in verblassterem sinne an die bedeutung ‘wenn’ rührend oder in sie übergegangen

‘in a more faded sense, touching the sense ‘if’ or having gone over to it’

In Present-Day German, the central sense (a) is now obsolete, judging by more recent dictionaries, which only give the conditional sense:

sofern ‘vorausgesetzt, dass [‘provided that’]’
e.g., Wir kommen morgen, sofern es euch passt (Duden 1999: s.v. sofern)

‘We’ll come tomorrow provided it suits you.’

Similar tendencies are found in the use of German soweit and Swedish såvitt ‘so far as; provided’. [11] More importantly, perhaps, the same cline is at work in one of the Finno-Ugric languages. Finnish mikäli (< mikä ‘what’), originally ‘in which direction, how far, by what route’ (Häkkinen 2007 s.v. mikäli), [12] developed degree and, more recently, conditional meanings as well (NSSK s.v.):

(31)

Mikäli kielitiede ja muinaistutkimus ovat pystyneet selvittämään Suomen suvun alkuperää, se näyttää johtavan Kaakkois-Venäjälle

‘As far as linguistics and research into the distant past have been able to clarify the origin of the Finnish nation, it seems to point to South-Eastern Russia’

(32)

Mikäli teosta on vielä saatavissa, pyydämme toimittamaan tilauksen ensi tilassa

‘If the work is still available, we request you to dispatch the order as soon as possible’

3. Conclusions

To our knowledge, the present paper is the first detailed study of the Old English adverbial subordinator be þæm þe/þæt. Why?

In our view we need not look far for the reasons. Be þæm þe is an item that shows extensive polyfunctionality in its semantics and syntax. Semantically, it either reflects some of the grammaticalized senses of the preposition be or, alternatively, relations that have arisen through grammaticalization processes within the subordinator. Syntactically, this unit has fuzzy borders, making it hard to distinguish it from prepositional be þæm phrases followed by subordinate clauses introduced by þe or þæt. This semantic iridescence and syntactic fuzziness, which of course are not unique to be þæm þe among the Old English prepositional subordinators, evidently contribute to the low level of salience of the subordinator, particularly as it is a relatively infrequent item. It only becomes practicable to study an elusive item like be þæm þe once all the evidence has become available in electronic form.

It is typical of be þæm þe that the various semantic relations it expresses do not reside in the unit itself, which has basically only deictic reference (cf. the discussion of forþæm in Lenker 2007: 207–209). The question of which relation applies to a particular instance must be answered by studying the co-text. Without the native speaker’s competence, the answers must necessarily remain tentative in many cases.

We have been able to establish seven adverbial relations for be þæm þe. Of these, most are mentioned in the literature (see 1.1 and 1.2), but the semantic relation of condition has not been recognized by the DOE or previous studies. This relation is extremely interesting from the point of view of the grammaticalization patterns of be þæm þe: we see it as the result of a final step within the CCCC domain from degree to condition and support our hypothesis with evidence from Finnish and German.

Using Kortmann’s seminal discussion of semantic affinities within the domain of adverbial subordination (1997: 197–204, 210), we would like to conclude this article by proposing a cognitive map of the semantic space of be þæm þe.

Notes

[1] There are no attestations of be þæm þe in either BT or BTS, s.vv. be and se.

[2] This study is an elaboration of the pilot work we presented at the Connectives in Synchrony and Diachrony in European Languages symposium. The pilot study was based on the York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE), which yielded only about 40 instances of be þæm þe.

[3] The spelling variants by and big are not attested within the be þæm þe phrase.

[4] To be precise, Kortmann’s 1997 discussion refers to this domain as CCC (cause, condition, concession, and related relations), with contrast being categorized as one of the ‘related relations’. However, since the publication of Couper-Kuhlen & Kortmann’s Cause, Condition, Concession, Contrast volume in 2000, the term ‘CCCC domain’ has become widely accepted instead of the earlier ‘CCC’.

[5] All the translations of the examples are by the authors.

[6] Judging by this comma Jost did not interpret be þæm þe here as one unit. We, however, find it difficult to accept his punctuation.

[7] We follow Halliday’s 1961 definition of a collocation as ‘the syntagmatic association of lexical items, quantifiable, textually, as the probability that there will occur, at n removes (a distance of n lexical items) from an item x [i.e. be þæm þe], the items a, b, c, …’ (1961: 276).

[8] The fact that adverbial finite comment clauses are often ‘cliché expressions’ is also observed by Quirk et al. (1985: §§15.54–55).

[9] See how medieval prefaces work towards the same end in Timofeeva (2006).

[10] Note further the univerbation of OHG bī thiu as bithiu (bidiu, bidhiu) in the examples cited in Wolfrum 1960, cf. example (13) in section 2.1.3.

[11] Ursula Lenker and Hanna Lehti-Eklund, personal communication.

[12] The original locative senses of mikäli are attested in Karelian and Ingrian. The word entered standard Finnish from eastern dialects. Matti Kilpiö’s idiolect recognizes only the conditional sense. For the degree sense he uses the expressions sikäli kuin, which contains the demonstrative correlative pair of mikäli, or siinä määrin kuin ‘to the extent that’.

References

Braune, Wilhelm & Ernst A. Ebbinghaus. [1880.] 1973. Gotische Grammatik. 18th edition. Tübingen: Niemeyer.

Brinton, Laurel J. 2008. Comment Clause: Syntactic Origins and Pragmatic Development. Cambridge: Cambridge University Press.

BT = Bosworth, Joseph and T. Northcote Toller. [1898] 1964. An Anglo-Saxon Dictionary. Oxford: Oxford University Press.

BTS = Bosworth, Joseph. [1921] 1955. An Anglo-Saxon Dictionary. Supplement by T. Northcote Toller. Oxford: Oxford University Press.

Couper-Kuhlen, Elizabeth & Bernd Kortmann, eds. 2000. Cause-Condition-Concession-Contrast: Cognitive and Discourse Perspectives. (=Topics in Linguistics, 33.) Berlin/New York: Mouton de Gruyter.

DOE = The Dictionary of Old English in Electronic Form on CD-ROM (2003), ed. by Antonette diPaolo Healey. Toronto: Toronto University Press.

DOEC = The Dictionary of Old English Corpus in Electronic Form (2004), ed. by Antonette diPaolo Healey, Toronto: University of Toronto, http://quod.lib.umich.edu/o/oec/.

Duden. Das große Wörterbuch der deutschen Sprache. 1999. 3rd, revised and expanded edition. Mannheim-Leipzig-Wien-Zürich: Dudenverlag.

Eearly English Laws website. Institute of Historical Research, King's College, London. Available online at http://earlyenglishlaws.as.uk. [Accessed: 14 December 2016].

Grimm, Jacob & Wilhelm Grimm. [1854] 1905. Deutsches Wörterbuch. Leipzig: Hirzel.

Häkkinen, Kaisa. 2007. Nykysuomen etymologinen sanakirja. Helsinki: WSOY.

Halliday, M.A.K. 1961. “Categories of the theory of grammar”, Word 17(3): 241–292.

Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization. 2nd edition. (=Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press.

Itkonen, Esa. 2002. “Grammaticalization as an analogue of hypothetico-deductive thinking”. New Reflections on Grammaticalization (=Typological Studies in Language, 49), ed. by Ilse Wischer & Gabriele Diewald, 413–422. Amsterdam and Philadelpia: Benjamins.

Kivimaa, Kirsti. 1966. Þe and Þat as Clause Connectives in Early Middle English with Especial Consideration of the Emergence of the Pleonastic Þat. (=Commentationes Humanarum Litterarum 39/1.) Helsinki: Societas Scientiarum Fennica.

Kortmann, Bernd. 1997. Adverbial Subordination. A Typology and History of Adverbial Subordinators Based on European Languages. (=Empirical Approaches to Language Typology, 18.) Berlin/New York: Mouton de Gruyter.

Lenker, Ursula. 2007. “Forhwi ‘because’: Shifting deictics in the history of English causal connection”. Connectives in the History of English (=Current Issues in Linguistic Theory, 283), ed. by Ursula Lenker & Anneli Meurman-Solin, 193–227. Amsterdam/Philadelphia: Benjamins.

Lenker, Ursula. 2010. Argument and Rhetoric: Adverbial Connectors in the History of English. (=Topics in English Linguistics, 64.) Berlin/New York: De Gruyter Mouton.

MED = The Electronic Middle English Dictionary. Ann Arbor: The University of Michigan Press, http://quod.lib.umich.edu/m/med/.

Mitchell, Bruce. [1985] 1987. Old English Syntax, 2 vols. Oxford: Clarendon Press.

NED = Murray, James A. H. (ed.). 1888. A New English Dictionary on Historical Principles, vol. 1: A-B. Oxford: Clarendon Press.

NSSK = Nykysuomen sanakirja. 1966. Ed. by Matti Sadeniemi et al. Porvoo/Helsinki: SKS.

OED = The Oxford English Dictionary, Online edition. 2008. London: Oxford University Press. http://www.oed.com/.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1972. A Grammar of Contemporary English. London: Longman.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London/New York: Longman.

Rissanen, Matti. 1997. “Optional THAT with subordinators in Middle English”. Language History and Linguistic Modelling: A Festschrift for Jacek Fisiak on his 60th Birthday. (Trends in Linguistics, Studies and Monographs, 101), ed. by Raymond Hickey & Stanislaw Puppel, 373–383. Berlin: Mouton de Gruyter.

Rissanen, Matti. 2007. “The development of adverbial subordinators in early English”. Change in Meaning and the Meaning of Change: Studies in Semantics and Grammar from Old to Present-Day English, ed. by Matti Rissanen, Leena Kahlas-Tarkka, Marianna Hintikka & Rod McConchie, 173–210. Helsinki: Société Néophilologique.

Rissanen, Matti. Forthcoming. “On the occurrence and variation of the adverbial subordination markers þe and þæt in Old English texts”.

Salminen, Taru. 2002. “Retention of abstract meaning: The essive case and grammaticalization of polyphony in Finnish”. New Reflections on Grammaticalization. (=Typological Studies in Language, 49), ed. by Ilse Wischer & Gabriele Diewald, 293–307. Amsterdam and Philadelpia: Benjamins.

Timofeeva, Olga. 2006. “Word be worde – andgit of andgite: A study of the medieval rhetorical formula”. Selected Proceedings of the 2005 Symposium on New Approaches in English Historical Lexis (HEL-LEX), ed. by R.W. McConchie et al., 135–142. Somerville, MA: Cascadilla Proceedings Project. http://www.lingref.com/cpp/hel-lex/2005/abstract1354.html.

Whitelock, Dorothy.1981. History, Law and Literature in 10th–11th Century England. London: Variorum Reprints.

Wolfrum, Gerhard. 1960. “Syntaktische Studien zu ahd. bī thiu”. Beiträge zur Geschichte der deutschen Sprache und Literatur 82: 226–241.

YCOE = The York-Toronto-Helsinki Parsed Corpus of Old English Prose, compiled by Ann Taylor, Anthony Warner, Susan Pintzuk & Frank Beths. University of York, http://www-users.york.ac.uk/~lang22/YcoeHome1.htm.

Appendix: DOEC texts containing be þæm þe/þæt

[UPDATE December 21, 2016 - LawEGu has been erroneously classified in the original version of this study as representing early OE. This item, however, belongs to the category of late OE texts (see Whitelock 1981 and http://www.earlyenglishlaws.ac.uk/laws/texts/egu). Due corrections have accordingly been made in Section 2.8, the Appendix and the list of references. MK and OT]

NB. The boldfaced texts represent Early Old English (= pre-950).

Total size of texts providing instances of be þæm þe/þæt: 367 414 words. Of these

• early OE: 51 444 words
• late OE: 315 970 words

Source for the word count: The Dictionary of Old English Corpus in Electronic Form, 2009 Release: November 2009.